Remember meForgot password?
    Log in with Twitter

article imageQ&A: Is it best to centralize or decentralize AI expertise? Special

By Tim Sandle     Aug 21, 2019 in Business
As businesses expand there artificial intelligence projects, should enterprises be centralizing or decentralizing AI? This is not a straightforward choice, as Albert Brown of Veritone explains.
A recent Gartner study found that 59 percent of organizations have deployed AI in their processes today and have on average four projects in the works at any one time. Going forwards, the typical large company has plans to add six more AI projects within the next year and another 15 within three years.
One thing driving AI project growth is the new wave of low-code tools on the horizon that will allow users without technical AI expertise to leverage the technology, leading to even more widespread adoption. But is this always the right way to go about AI initiatives? Or should enterprises keep AI centralized for better results? Albert Brown, SVP of Engineering at Veritone (NASDAQ: discusses whether centralizing or decentralizing AI is the better choice.
Digital Journal: How rapidly are businesses adopting AI?
Albert Brown: Businesses are beginning to recognize that AI has the potential to radically grow their business. A few years ago, only a handful of early adopters were using AI and it was being leveraged for primarily research-based projects. But as we move out of this early adoption phase, we are entering a new stage where AI is being used to solve actual business problems. For example, companies have baked AI into their office security procedures and buildings integrate facial recognition technology into their security processes to unlock doors, recognize who has clearance for specific areas and approve entrance.
DJ: What are some of the main tasks that AI is being used for in businesses?
Brown: Originally, companies used AI mainly for prediction and classification of structured information. Today businesses are embedding AI into a number of their processes, mostly to reduce time and resources for mundane, repetitive tasks including face recognition of customers, employees and persons of interest as well as compliance-based tasks.
For example, now, AI has the opportunity to boost compliance across organizations. In the past, businesses used random sampling to identify fraud and set compliance standards. Businesses are starting to leverage AI to provide complete coverage, rather than listening to two or three percent of trader calls. The transcription and translation powered by AI automatically flag information that might need reviewing, meaning AI strengthens businesses compliance standards without added time for knowledge workers.
DJ: What will the consequences be of the entry of low-code AI tools to the marketplace?
Brown: As recently as a couple of years ago, AI was still largely in its early adopter phase, where data scientists created and experimented with building AI-solutions. Now, low-code tools like RPA directly put the technology into knowledge workers’ hands, and users can apply it to their own tasks. Armed with these tools, employees are able to automate more routine tasks and instead spend the bulk of their time on things that require actual human intelligence. As a result, the workplace is changing – especially IT departments.
Consider the digital office, where IT is building personalized, data-specific camera networks to automatically check-in visitors. Before low-code tools, companies needed data scientists to build a specific solution for the organization to use in these situations. Now, the IT department has the ability to personalize a specific application for their unique needs – without the extra help or expertise from a data scientist. As a result, organizations are able to build solutions in weeks, rather than months – saving time and money while increasing efficiency.
DJ: Should enterprises be keeping AI centralized or letting it go out to different divisions?
Brown: In the early stages of development and adoption within organizations, AI needs to be centralized. Those who specialize in AI need to fully understand its capabilities and specific use cases, then build out its core competencies to identify how to use the technology throughout the organization, before it’s more widely adopted by the company.
To start, establish a center of excellence for AI, with the goal to understand how to use AI in the business at a very high level. Then, organizations can drill down to specific departments with a blueprint and guideline that outlines how each could take advantage of AI in different ways. When businesses start by decentralizing their AI expertise, these processes are extremely fragmented, meaning that different departments will be able to do what they individually want.
While this may be great for each department, it’s ultimately detrimental to the organization as a whole since the collective shared business goal is not defined. To successfully adopt AI, I recommend that organizations begin with their AI expertise centralized, then slowly extend it to different departments once the larger business goals and guidelines are set. Most importantly, define your goals before you start; it is all too easy to get distracted along the way.
DJ: Should companies put AI in the hands of all their employees?
Brown: Before putting AI in the hands of their employees, organizations must first understand it – just like any other complex system or technology. A group of AI specialists need to identify how the business can use AI, then build organization-specific guidelines for the rest of the organization to follow, before it’s more widely-adopted for specific, individual purposes.
Current low-code tools, like RPA, make it easier to embed AI into existing business processes at a faster speed and enable a wider user base because they don’t need as strong of an understanding of the technology. That said, it’s important that organizations themselves understand how to effectively use AI at a higher level before putting the technology in the hands of all employees through these low-code tools which I believe will ultimately enable and accelerate AI adoption across the organization.
DJ: What developments can we expect from AI over the next five years?
Brown: As AI adoption grows, the structure of businesses will look different. For instance, organizations will need to hire Chief Data Officers to lead the transformation of their organizations into data-driven companies and their AI adoptions. Data gives organizations the information they need to make decisions and target the business problems that AI can solve. When organizations recognize the value of data, they will work more cross-functionally to prioritize data collection and analysis. This won’t take long for the digitally native organizations, but more traditional businesses that have been around for decades will need to dedicate more time automating processes, collecting data and targeting desired outcomes.
More about Artificial intelligence, decentralized systems, Enterprise
More news from
Latest News
Top News